代码之家 › 专栏 › 技术社区 › Mohan

如何使用任意函数转换数据帧中的每一行?

rowwise dplyr r

0

Mohan · 技术社区 · 10 月前

我有一个数据帧,其中每一行都包含特定年龄(0、6、15、24、36个月)的值。我可以使用以下公式将一行的值插值到其他年龄段 approx 。我如何将此应用于所有行?我尝试过各种各样的东西 rowwise() ,但它们不起作用。

library(tidyverse)
library(stats)

care_distribution_table = "Age,M0,M6,M15,M24,M36
Mother,100.00%,36.89%,29.58%,29.61%,26.64%
Father,0.00%,11.74%,14.70%,12.23%,11.40%
Other relative,0.00%,15.80%,14.00%,11.55%,10.20%
In-home nonrelative,0.00%,7.94%,8.89%,7.33%,6.48%
Home-based childcare,0.00%,18.62%,20.69%,21.82%,17.38%
Center childcare,0.00%,9.00%,12.15%,17.46%,27.90%"

care_dist = read.csv(text = care_distribution_table, row.names = 1) |>
  # Convert from % to float
  mutate(across(everything(), ~as.numeric(trimws(.x, whitespace = '%'))) / 100) |>
  # Make sure columns add to 1 (i.e. handle rounding error)
  mutate(across(everything(), ~.x/sum(.x))) 

# This interpolates one row
approx(x = c(0,6,15,24,36), y = care_dist['Mother',], xout = 0:36)


# This gets me most of the way
care_dist |> rowwise() |> group_map(~approx(x = c(0,6,15,24,36), y = .x, xout = 0:36)$y) 

# This fails
care_dist |> rowwise() |> group_modify(~df(z = approx(x = c(0,6,15,24,36), y = .x, xout = 0:36)$y))

2 回复 | 直到 10 月前

1

2

Ben Bolker 10 月前

如果 CC 是您的输出 group_map() 呼叫,然后

bind_cols(care_dist, do.call(rbind, CC))

将为您提供一个包含所有近似值(带名称)的宽格式tibble ...6 , ...7 , ... 也许你想写一个包装器函数来给这些更好的名字?)

(我不知道为什么我需要 do.call(rbind, ...) 而不是 bind_rows() .)

2

G. Grothendieck 10 月前

假设所需的输出是一个具有原始行名和M*列名的6 x 37数据帧

library(dplyr)
library(tibble)

care_dist %>%
  rownames_to_column %>%
  reframe(
    approx(x = c(0,6,15,24,36), y = unlist(pick(everything())), xout = 0:36)$y %>%
      as.data.frame.list(col.names = paste0("M", 0:36)), .by = rowname
  ) %>%
  column_to_rownames